Optimizing Collective Pitch Control in Wind Turbines Using Deep Reinforcement Learning and OpenFAST Simulation

The increasing global reliance on wind energy requires advanced control strategies to enhance wind turbine efficiency and lifespan. This study develops a reinforcement learning-based collective pitch controller for above-rated wind conditions. This research employs a Double Deep Q-Network approach integrated with OpenFAST using a novel two-stage methodology: policy transfer, which extracts an initial policy from a standard controller, and policy refinement, where the reinforcement learning agent optimizes its strategy. The proposed controller outperforms the industry-standard ROSCO controller, achieving a 15% reduction in power fluctuations, demonstrating the potential of reinforcement learning-based control to enhance wind turbine performance.